Printed circuit boards (PCBs), laminate chip carriers, and the like permit formation of multiple circuits in a minimum volume or space. Such structures typically comprise a stack of layers of signal, ground and/or power planes (lines) separated from each other by a layer of dielectric material. The lines on one plane are often in electrical contact with those on another plane by plated holes passing through the dielectric layers. The plated holes are often referred to as “vias” if internally located, “blind vias” if extending a predetermined depth within the board from an external surface, or “plated-thru-holes” (PTHs) if extending substantially through the board's full thickness. By the term “thru-hole” as used herein is meant to include all three types of such board openings, or “holes.”
Known methods for fabricating PCBs, chip carriers and the like typically comprise fabrication of separate inner-layer circuits (circuitized layers), which are formed by coating a photosensitive layer or film over a copper layer of a copper clad inner-layer base material. The photosensitive coating is imaged, developed and the exposed copper is etched to form the desired number of conductor lines. The photosensitive film is then stripped from the copper, leaving the circuit pattern on the surface of the inner-layer base material. Often, this methodology is referred to as photolithographic processing in the PCB art. Added description is not believed necessary in view of such known teachings.
A multilayered stack is formed, following formation of the desired number of inner-layer circuits, by preparing a lay-up of inner-layers, ground planes, power planes, etc., typically separated from each other by a layer of dielectric “pre-preg” material, which usually includes a layer of glass cloth (e.g., fiberglass) impregnated with a partially cured material (e.g., a “B-stage” epoxy resin). The outermost (top and bottom) layers of the stack usually comprise copper clad, glass-filled, epoxy planar substrates with the copper cladding comprising exterior surfaces of the stack. This stack is laminated to form a monolithic structure using heat and pressure to fully cure the B-stage resin. As understood, the resulting stack typically has metal (usually copper) cladding on both of its exterior surfaces. Exterior circuit layers are formed in the copper cladding using procedures similar to the procedures used to form the inner-layer circuits. In a well known such procedure, a photosensitive film is applied to the copper cladding, exposed to patterned activating radiation, and developed. An etchant is then used to remove copper bared by the development of the photosensitive film. Finally, the remaining photosensitive film is removed to provide the exterior circuit layers. Various elements of these outer layers, such as pads, may then be electrically coupled to selected electronic components mounted on the structure, such components including capacitors, resistors, modules, and the like, including even semiconductor chips.
Conductive thru-holes (or “interconnects”, as often referred to in the art) are used to electrically connect individual circuit layers within the structure to each other and/or to the outer surfaces, these thru-holes passing through all or a portion of the “stack”. Thru-holes are generally formed prior to the formation of circuits on the exterior surfaces by drilling holes through the stack at appropriate locations. Following several pre-treatment steps, the walls of the holes are typically catalyzed by contact with a plating catalyst and metallized, typically by contact with an electro-less or electrolytic copper plating solution to form conductive pathways between circuit layers. Following formation of the conductive thru-holes, exterior circuits, or outer conductive layers are formed using the above procedure(s).
Following the described construction, the aforementioned semiconductor chips and/or other electrical components are mounted at appropriate locations on the exterior circuit layers of the multilayered structure, typically using solder mount pads to bond the components. These components are usually in electrical contact with the circuits within the structure through the conductive thru-holes, as desired. Such solder pads are typically formed by coating an organic solder mask coating over the exterior circuit layers. The solder mask may be applied by screen coating a liquid solder mask coating material over the surface of the exterior circuit layers using a screen having openings defining areas where solder mount pads are to be formed. Alternatively, a photoimageable solder mask may be coated onto the board and then exposed and developed to yield an array of openings defining the pads. The openings are then coated with solder using processes known to the art, one known process being wave soldering.
Significantly, the relative complexity of the designs of resulting products as described herein has increased substantially over the past few years. Mainframe computer PCBs, for example, may require as many as thirty-six layers of circuitry or more, with the complete structure having a thickness of as much as about 0.250 inch (250 mils). These boards are typically designed with about three or five mil wide signal lines and twelve mil diameter thru-holes. For increased circuit densification in many of today's products such as PCBs, chip carriers and the like, the industry desires to reduce signal lines to a width of two mils or less and thru-hole diameters to two mils or less. Most known commercial procedures, especially those of the nature described herein, are incapable of economically forming the dimensions desired by the industry in a cost effective manner.
The PCB industry, in addition to requiring decreasing line widths and via diameters, also desires to avoid manufacturing problems frequently associated with their PCB products, chip carriers and the like. As stated, current processes utilize inner-layer materials that are glass-reinforced resin or other suitable dielectric material layers having a thickness of from about two to five mils, clad with metal (typically copper) on both surfaces. Glass-reinforcing material, typically utilizing continuous strands of fiberglass which extend throughout the width and length of the overall final substrates, is used to contribute strength and rigidity to the final stack. Being continuous, these strands commonly run the full width (or length) of the structure and include no breaks or other segments as part thereof. Thus, by the term “continuous” as used herein to define fibrous materials is meant a structure such as a woven cloth of lengthy fibers, including fibers which, as stated, typically run the full distance through the structure. By the term “semi-continuous” as used herein (below) to define fibrous materials is meant structures with much shortened length fibers, which are also referred to as “chopped” fibers, such as chopped fiber mats. Such fibrous materials occupy a relatively significant portion of the substrate's total volume, a disadvantage especially when attempting to produce highly dense numbers of thru-holes and very fine line circuitry to meet new, more stringent design requirements. More specifically, when holes are drilled (typically using laser or mechanical drills) to form these needed thru-holes, end segments of the fiberglass fibers may extend into the holes during lamination, and, if so, must be removed prior to metallization. This removal, in turn, creates the need for additional pretreatment steps such as the use of glass etchants to remove the glass fibrils extending into the holes, subsequent rinsing, etc. If the glass is not removed, a loss of continuity might occur in the hole's internal wall metal deposit. In addition, both continuous and semi-continuous glass fibers add weight and thickness to the overall final structure, yet another disadvantage associated with such fibers. Additionally, since lamination is typically at a temperature above 150° C., the resinous portion of the laminate usually shrinks during cooling to the extent permitted by the rigid copper cladding, which is not the case for the continuous strands of fiberglass or other continuous reinforcing material used. The strands thus take on a larger portion of the substrate's volume following such shrinkage and add further to complexity of manufacture in a high density product. If the copper is etched to form a discontinuous pattern, laminate shrinkage may not be restrained even to the extent above by the copper cladding. Obviously, this problem is exacerbated as feature sizes (line widths and thicknesses, and thru-hole diameters) decrease. Consequently, even further shrinkage may occur. The shrinkage, possibly in part due to the presence of the relatively large volume percentage of continuous or semi-continuous fiber strands in the individual layers used to form a final product possessing many such layers, may have an adverse affect on dimensional stability and registration between said layers, adding even more problems for the PCB manufacturer.
The presence of glass fibers, especially woven glass fibers, also substantially impairs the ability to form high quality, very small thru-holes using a laser. Glass cloth has drastically different absorption and heat of ablation properties than typical thermo-set or thermo-plastic matrix resins. In a typical woven glass cloth, for example, the density of glass a laser might encounter can vary from approximately 0% in a window area to approximately 50% by volume or even more, especially in an area over a cloth “knuckle”. This wide variation in encountered glass density leads to problems obtaining the proper laser power for each thru-hole and may result in wide variations in thru-hole quality, obviously unacceptable by today's very demanding manufacturing standards.
The presence of glass fibers in a multilayered structure of the types mentioned herein also often contributes to an electrical failure mode known as CAF growth. CAF (cathodic/anodic filament) growth often results in an electrical shorting failure which occurs when dendritic metal filaments grow along an interface (typically a glass fiber/epoxy resin interface), creating an electrical path between two features which should remain electrically isolated. Whether continuous (like woven cloth) or semi-continuous (like chopped fiber mattes), glass fiber lengths are substantial in comparison to the common distances between isolated internal features, and thus glass fibers can be a significant detractor for PCB insulation resistance reliability. While the use of glass mattes composed of random discontinuous chopped fibers (in comparison to the longer fibers found in continuous structures) can largely abate the problem of inadequate laser drilled thru-hole quality, such mattes still contain fibers with substantial length compared to internal board feature spacing and, in some cases, offer virtually no relief from the problem of this highly undesirable type of growth.
In U.S. Pat. No. 5,246,817, there is described one form of improvement in the manufacture of products such as PCBs. The manufacturing process in U.S. Pat. No. 5,246,817 consists of the sequential formation of layers using photosensitive dielectric coatings and selective metal deposition procedures. The first layer of the board is formed over a temporary or permanent carrier, which, if the latter, will become an integral part of the board. When the carrier is a circuit, the process comprises formation of a dielectric coating over the circuit with imaged openings defining the thru-holes. The imaged openings may be formed by exposure of a photosensitive dielectric coating to activating radiation through a mask in an imaged pattern, followed by a described development procedure. Alternatively, imaging may be by laser ablation, in which case, the dielectric material need not be photosensitive. Metal is deposited into the recesses within the dielectric coating to form the conductive thru-holes. Thereafter, an additional layer of dielectric material is coated onto the first dielectric layer, imaged in a pattern of circuit lines, and the recesses are then plated with metal. After imaging the first dielectric coating, it may alternatively be coated with a second dielectric coating and imaged and the recesses plated with metal to form the thru-holes and circuit lines simultaneously. By either process, the walls of the imaged opening or recesses in the dielectric coating contain metal as it deposits during plating and assures a desired cross-sectional shape of the deposit. Plating desirably fills the entire recess within the imaged photosensitive coating. The process, obviously very complex and costly, is repeated sequentially to form sequential layers of circuits and thru-holes.
In U.S. Pat. No. 6,207,595, there is described another example of a dielectric material composition for use in a PCB in which the dielectric layer's fabric material is made from a cloth member having a low enough content of particulates and a sufficient quantity of resin material to completely encase the cloth member including the particulates, so that the resin material extends beyond the highest protrusions of the cloth member (i.e. the fabric material is thicker and will pass a certain test standard (in '595, the known HAST level A test). Thus, the woven cloth is known to include a quantity of particulates, which term is meant in '595 to include dried film, excess coupler, broken filaments, and gross surface debris. A process is described where a sizing of polyvinyl alcohol, corn starch and a lubricant of oil is applied to the strands of fiber prior to weaving in order to improve the weaving process and minimize breakage of the strands. After weaving, the sizing is removed by a firing step to clean the filaments of lubricants and other materials. However, some sizing is randomly left behind as particulates. Encasing the woven cloth including the particulates is a quantity of hardened resin material. The resin may be an epoxy resin such as one often used for “FR4” composites (“FR4” has become a conventional, abbreviated name for the resulting substrates and often also for the resins forming part thereof, and is based in part on the flame retardant (hence the “FR” designation) rating of these established products). A resin material based on bismaleimide-triazine (BT) is also acceptable for the structure in this patent. More preferably, the resin is a phenolically hardenable resin material known in the PCB industry. This patent thus requires continuous fibers (those extending across the entire width (or length) of the dielectric layer except for possible inadvertent interruptions caused by drilling of the thru-holes needed in the final product, causing these fibers to become what might be called as “broken.” The aforementioned problem with fiber strands exposed to the holes is thus possible in this patent's process and resulting structure.
In U.S. Pat. No. 5,418,689, there is described a PCB product wherein the dielectric substrate can include a thermoplastic and/or thermosetting resin. Thermosetting polymeric materials mentioned in this patent include epoxy, phenolic base materials, polyimides and polyamides. Examples of some phenolic type materials include copolymers of phenol, resorcinol, and cresol. Examples of some suitable thermoplastic polymeric materials include polyolefins such as polypropylene, polysulfones, polycarbonates, nitrile rubbers, ABS polymers, and fluorocarbon polymers such as polytetrafluoroethylene, polymers of chlorotrifluoroethylene, fluorinated ethylenepropylene polymers, polyvinylidene fluoride and polyhexafluoropropylene. The dielectric materials may be molded articles of the polymers containing fillers and/or reinforcing agents such as glass filled polymers. “FR4” epoxy compositions that are employed in this patent contain 70-90 parts of brominated polyglycidyl ether of bisphenol-A and 10-30 parts of tetrakis(hydroxyphenyl)ethane tetraglycidyl ether cured with 3-4 parts of dicyandiamide, and 0.2-0.4 parts of a tertiary amine, all parts being parts by weight per hundred parts of resin solids. Another “FR4” epoxy composition may contain about 25 to about 30 parts by weight of a tetrabrominated digylcidyl ether of bisphenol-A having an epoxy equivalent weight of about 350 to about 450; about 10 to about 15% by weight of a tetrabrominated glycidyl ether of bisphenol-A having an epoxy equivalent weight of approximately 600 to about 750 and about 55 to about 65 parts per weight of at least one epoxidized, nonlinear novolak having at least 6 terminal epoxy groups; along with suitable curing and/or hardening agents. A still further “FR4” epoxy composition contains 70 to 90 parts of brominated polyglycidyl ether of bisphenol-A and 10 to 30 parts of tetrakis(hydroxyphenyl)ethane tetraglycidyl ether cured with 0.8-1 phr of 2-methylimidazole. Still other “FR4” epoxy compositions employ tetrabromobisphenol-A as the curing agent along with 2-methylimidazole as the catalyst.
In U.S. Pat. No. 6,323,436, PCBs are prepared by first impregnating a non-woven aramid chopped fiber mat or a thermoplastic liquid crystalline polymer (LCP) paper instead of the reinforcement typically used in the electronics industry, described in this patent as a woven glass fabric. This aramid reinforcement matte is comprised of a random (in-plane) oriented mat of p-aramid (poly(p-phenylene terephthalamide)) fibers comprised of Kevlar (Kevlar is a registered trademark of E. I. duPont deNemours and Company), and has a dielectric constant of 4.0 as compared to 6.1 for standard E-glass cloth. The lower permittivity of the non-woven aramid reinforcement provides for faster signal propagation, allowing increased wiring density and less crosstalk, which becomes increasingly important for high I/O chips and miniaturization. Since the p-aramid fibers are transversely isotropic and have an axial CTE of about −3 to about −6 ppm/degree Celsius (hereinafter C.) when combined with a thermosetting resin, the final composite described in this patent is said to possess a CTE which can be controlled and adjusted to match that of silicon or semiconductor chips in the range of about 3 to about 10 ppm/degree C. The thermoplastic liquid crystal polymer (LCP) paper is a material called Vecrus (Vecrus is a registered trademark of Hoechst Celanese Corp.), which uses the company's Vectra polymer as part thereof (Vectra also being a registered trademark of Hoechst Celanese Corp.). According to this patent, the paper has a dielectric constant of 3.25, a dissipation factor of 0.024 at 60 Hertz (Hz), a UL 94-V0 rating and an in-plane CTE of less than 10 ppm/degree. C. The alleged advantages of this material over the aramid mat are the lower dielectric constant and very low moisture absorption, allegedly less than 0.02%. The non-woven aramid or LCP paper is used in conjunction with a thermosetting resin to form the final composite substrate. Examples of thermosetting resins useful in this patent include epoxy, cyanate ester, bismaleimide, bismaleimide-triazine, maleimide or combinations thereof. The resin-impregnated low CTE reinforcement is then partially cured to a “B”-stage to form the pre-preg, and then the pre-preg is cut, stacked, and laminated to form a sub-composite with exterior copper sheets.
In U.S. Pat. No. 5,314,742, there is described the use of non-woven aramid sheets to provide reinforcement for the resulting laminate. The reinforcing aramid sheet is described as having a coefficient of thermal expansion (CTE) of less than 10 ppm per degree. C and is prepared from 75 to 95 wt. % p-aramid floc and from 5 to 25 wt. % poly(m-phenylene isophthalamide) fibrids. Floc is defined in U.S. Pat. No. 4,729,921. Para-aramid fibers are very high in strength and modulus. Examples of para-aramid fibers are set out in U.S. Pat. No. 3,869,429. Specific examples of para-aramid materials are poly(p-phenylene terephthalamide) (PPD-T) and copoly(p-phenylene-3,4′-oxydiphenylene terephthalamide). Fibers of PPD-T are generally made by an air gap spinning process such as described in U.S. Pat. No. 3,767,756, and are preferably heat treated as described in U.S. Pat. No. 3,869,430. Preferably, poly (p-phenylene terephthalamide) floc which has not been refined is utilized. High shear forces exerted on the fibers during processing, e.g., refining, may cause damage to the fibers and adversely affect the CTE of the reinforcement. It is also preferred to employ p-aramid floc of high orientation and relatively lower crystallinity. Fibrids are described in U.S. Pat. No. 4,729,921. To prepare the sheet, the floc and fibrids are dispersed in the desired proportions as an aqueous slurry, the solids concentration generally ranging between 0.005% and 0.02%. The slurry is not refined. The slurry can be made into paper by conventional means. In the examples mentioned in this patent, wet sheets were formed in an inclined wire Deltaformer papermaking machine and dried using heated drier cans. The dried sheets preferably had a basis weight between 0.8 and 4.0 oz/yd2, and were then calendered between two hard-surface rolls. Calender pressures between about 500 and 2500 kg/cm (nip pressure) and roll temperatures between about 130 and 150 degrees C. were used. The paper was then pre-pregged with a resin having a high glass transition temperature (Tg), e.g., above about 160 degrees C.
“Expanded PTFE” dielectric materials represent another type of materials known for use in circuitized substrates, PTFE being the designate for polytetrafluoroethylene. A common example of such material is the aforementioned Teflon, sold by E. I. DuPont de Nemours and Company. In U.S. Pat. No. 5,652,055, for example, there is described an adhesive sheet (or “bond ply”) material suitable to serve as adhesive layers in a variety of adhesive applications, such as in circuit board laminates, multi-chip modules, and in other electrical applications. The adhesive sheet is described as being constructed from an expanded PTFE material, such as that described in U.S. Pat. No. 3,953,566. Preferably, the material is filled with inorganic filler and is constructed as follows: A ceramic filler is incorporated into an aqueous dispersion of dispersion-produced PTFE. The filler in small particle form is ordinarily less than 40 microns in size, and preferably less than 15 microns. The filler is introduced prior to co-coagulation in an amount that will provide 10 to 60%, and preferably 40 to 50% by weight filler in the PTFE, in relation to the final resin-impregnated composite. The filled PTFE dispersion is then co-coagulated, usually by rapid stirring. The coagulated filled PTFE is then added. The filled material is then lubricated with a common paste extrusion lubricant, such as mineral spirits or glycols, and then paste extruded. The extrudate is usually calendered, and then rapidly stretched 1.2 to 5000 times, preferably 2 times to 100 times, per this patent, at a stretch rate of over 10% per second, at a temperature of between 35 degrees C. and 327 degrees C. The lubricant can be removed from the extrudate prior to stretching, if desired. The resulting expanded, porous filled PTFE is then imbibed with adhesive by dipping, calendaring, or doctor blading on a varnish solution of about 2% to 70% adhesive in solvent. The wet composite is then affixed to a tenter frame, and subsequently “B-staged” at or about 165 degrees C. for 1 to 3 minutes. The resulting sheet adhesive typically consists of: (a) 9 to 65 weight percent PTFE; (b) 9 to 60 weight percent inorganic filler, in the form of particulate; and (c) 5 to 60 weight percent adhesive imbibed within the porous web structure.
Other types of expanded-PTFE substrate materials are described in the aforementioned U.S. Pat. No. 3,953,566, and also in U.S. Pat. Nos. 4,187,390 and 4,482,516, as well as many others. U.S. Pat. No. 4,187,390 is particularly interesting because it delves substantially into both nodes and fibrils used as part of such substrate materials, breaking these down into such dimensional constraints as node height, node width, node length, and fibril length.
Other examples of methods of making circuitized substrates such as PCBs are described and illustrated in the several documents cited in the aforementioned co-pending application Ser. No. 10/812,890, the teachings of which, in addition to those other patents and applications listed above, are incorporated herein by reference.
The several examples of dielectric materials mentioned in many of the above documents are considered to have a limited usefulness as a substrate dielectric because of what can be referred to as high moisture absorptivity. That is, these dielectric materials tend to absorb what might be referred to as undesirable levels of moisture (a primary example being water) during processing and subsequent storage thereof. This unacceptable moisture results in the dielectric structure having unpredictable dimensional stability during processing and delamination, or blistering in the final component or board assembly process. By the term “high moisture absorptivity” as used herein is meant a level of moisture (water and other fluid materials, including other processing liquids and even gases to which the dielectric material is exposed during processing, storage and shipping) absorption greater than about 0.27 percent, as measured at a temperature of about 22° C. for a time period of about 24 hours. A specific example used in the industry is when the dielectric layer is immersed in water at this temperature for this time period. An absorption level greater than about 0.27 percent is deemed unacceptable for the reasons given above.
The present invention as defined herein represents a significant improvement over products and processes such as those described above. One particularly significant feature of this invention is the provision of a composite layer including a first dielectric sub-layer including a plurality of fibers having a low coefficient of thermal expansion and a second dielectric sub-layer of a low moisture absorptivity resin not including continuous or semi-continuous fibers or the like as part thereof. The invention is thus uniquely able to combine the significant advantages of fiber use, especially those relating to reinforcement, with a non-fibrous second sub-layer including a low moisture absorptivity resin, to produce a multilayered structure in which high density arrays of thru-holes are possible without, substantially, the aforementioned disadvantages associated with such fiber usage. Low moisture absorptive polymer resin is defined herein as a resin that absorbs less than about 0.27 percent by weight moisture at room temperature (22° C.) for a period of 24 hours, while immersed in water.
It is believed that such an invention will represent a significant advancement in the art.